The objective of this investigation is to improve the forecasting accuracy of the Tadawul market data patterns by modeling daily return data using 4861 observations spanning from January 2005 to February 2024. The model combines five mathematical functions with an adaptive network-based fuzzy inference system (ANFIS) and a nonlinear spectral model that uses the maximal overlapping discrete wavelet transform (MODWT). Utilizing correlation, multiple regressions, the Autocorrelation Function (ACF), and the Partial Autocorrelation Function (PACF), the selected input values are the correlated first difference of return (D(return(− 1))), the second-order differencing of the return with a lag of 1 (D(return(− 1),2), the second-order differencing of the return with a lag of 2 (D(return(− 2),2), and the second-order differencing of the return with a lag of 4 (D(return(− 4),2). These factors were supplied by the Saudi Stock Exchange’s Tadawul market. The Autoregressive Featured Information System (ANFIS) model and the Autoregressive Integrated Moving Average (ARIMA) model are two classical models that are used to compare the performance of the proposed MODWT-bl14-ANFIS model. The results demonstrate that the MODWT-bl14-ANFIS model outperforms the traditional models. Consequently, the suggested prediction model offers a viable strategy that can be used successfully in stock exchange markets.